Hello. I used GLIMMIX to do the age-period-cohort analysis. CASE_NUM is number of cases. LOGPOP is Log(Number of base population). proc glimmix data=A ic=q; class AGE(ref='50') PERIOD COHORT; model CASE_NUM = AGE  /dist=poisson solution offset=LOGPOP; random PERIOD COHORT /solution; covtest glm/wald; NLOPTIONS TECHNIQUE=NRRIDG; ; run;   My question is as follows. 1> Using GLIMMIX with random variables like this can avoid identification problems, that is most big issues in APC analysis? 2> Can I find any examples using GLIMMIX for the APC analysis?   Regards.. ... View more